[jboss-jira] [JBoss JIRA] (JGRP-1886) CENTRAL_LOCK: provide option to make a node the lock owner instead of node:thread

[Bela Ban (JIRA)]

     [ https://issues.jboss.org/browse/JGRP-1886?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Bela Ban updated JGRP-1886:
---------------------------
    Description: 
When using distributed locks, we currently use {{Owner}} as identity that holds a lock. {{Owner}} consists of the node's address and the thread's ID.
This has the same semantics as {{ReentrantLock}} which makes the thread which called {{Lock.lock()}} the owner.
However, in some scenarios, this is too strong and some applications only want the current node to be the lock owner. This is needed in cases where thread T1 locks a lock but thread T2 needs to unlock it.
This means, however, that all threads in a given node can lock or unlock the same lock.
h4. Solution
* Add property {{use_thread_id_for_lock_owner}} to {{CENTRAL_LOCK}} (default: false)
* When set, {{getOwner()}} sets the thread-id to -1

  was:
When using distributed locks, we currently use {{Owner}} as identity that holds a lock. {{Owner}} consists of the node's address and the thread's ID.
This has the same semantics as {{ReentrantLock}} which makes the thread which called {{Lock.lock()}} the owner.
However, in some scenarios, this is too strong and some applications only want the current node to be the lock owner. This is needed in cases where thread T1 locks a lock but thread T2 needs to unlock it.
This means, however, that all threads in a given node can lock or unlock the same lock.
h4. Solution
* Add property {{use_thread_id_for_lock_owner}} to {{CENTRAL_LOCK}} (default: false)
* When set, {{getOwner()}} sets the thread-id to 0



> CENTRAL_LOCK: provide option to make a node the lock owner instead of node:thread
> ---------------------------------------------------------------------------------
>
>                 Key: JGRP-1886
>                 URL: https://issues.jboss.org/browse/JGRP-1886
>             Project: JGroups
>          Issue Type: Feature Request
>            Reporter: Bela Ban
>            Assignee: Bela Ban
>             Fix For: 3.6
>
>
> When using distributed locks, we currently use {{Owner}} as identity that holds a lock. {{Owner}} consists of the node's address and the thread's ID.
> This has the same semantics as {{ReentrantLock}} which makes the thread which called {{Lock.lock()}} the owner.
> However, in some scenarios, this is too strong and some applications only want the current node to be the lock owner. This is needed in cases where thread T1 locks a lock but thread T2 needs to unlock it.
> This means, however, that all threads in a given node can lock or unlock the same lock.
> h4. Solution
> * Add property {{use_thread_id_for_lock_owner}} to {{CENTRAL_LOCK}} (default: false)
> * When set, {{getOwner()}} sets the thread-id to -1



--
This message was sent by Atlassian JIRA
(v6.3.1#6329)